Attachment
Gates Package, p.1329 · gates:exh:00616
Page text: p.1329 · original PDF
- Date
- — (unknown precision)
- Type
- attachment · document
@) Preliminary
PREDICT AND DESIGN DE NOVO ODORS
Neural network
Molecule to graph operation
[> Neural network for prediction
I Fe
138 odor descriptions
Molecule
(eg, varilin)
citrus
creamy
I
sweet
baked -- spicy
HOW ARE BEHAVIOR-MODIFYING ODORANTS IDENTIFIED?
Learned odor space*
Scent prediction
Novel molecule discevery
for behavior modification
(e-g. insect repellant
Novel scent
odorless
varia
neue)
lean
alcoholic
beefy
network
checolate fruity
layers
Graph embeddings
Graph
bX)
>
to vector
operation LE
Tn TJ --
Can we redesign and retune olfactory
receptors to detect volatiles of human interest
and importance?
A
"This graph only shows two of the most important
dimensions in the odor space.
In realty, the odor space is > 100 dimensions.
Identify new behavior-modifying
chemicals and odors
WHAT POTENTIAL RELATIONSHIPS
COULD ACCELERATE THIS WORK?
TROPIG
SCJohnson
HEA as
SCIENCES
USDA
--
And more,
including new
companies
POSSIBLE APPLICATION(S)
New classes of mosquito repellants/
Cc)
attractants that protect humans/
animals from vector born infections
Not for distribution\ 1